SPE-169945-MS

A Model for Extending the Heptanes plus Fraction for Trinidad Gas
Condensates
R. Mayrhoo, and R. Hosein, The University of The West Indies

Copyright 2014, Society of Petroleum Engineers

This paper was prepared for presentation at the SPE Biennial Energy Resources Conference held in Port of Spain, Trinidad, 09 –11 June 2014.

This paper was selected for presentation by an SPE program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Society of Petroleum Engineers and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Society of Petroleum Engineers, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written
consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may
not be copied. The abstract must contain conspicuous acknowledgment of SPE copyright.

Abstract
An accurate description of pseudo-component compositions is required for Equation of State predictions
for gas condensate systems. Very often these extended experimental data are unavailable and must be
generated using mathematical models, of which the exponential and the three-parameter gamma distri-
bution functions are the two most widely used. The development of these two techniques was based on
the assumption of a continuous molar relationship for pseudo-components. However, experimental
compositional data for gas condensate systems show discontinuities in this relationship at SCN8 and
SCN13.
The above models when applied to extend the C7⫹ fraction for Trinidad gas condensates, under predict
the SCN8 mole percent and over predict the SCN12 mole percent due to the aforementioned disconti-
nuities. The Average Absolute Deviation between the predicted and experimental SCN8 and SCN12 data
were both greater than 25 percent. The two coefficient method described by Ahmed et al., when applied
to extend the C7⫹ fraction, reduced the discontinuity at SCN8 to less than 12 percent. However the SCN12
group still had a deviation greater than 18 percent. These results show that existing models were not
designed to take care of these discontinuities and should be used with caution when extending experi-
mental data beyond SCN7.
The Model described in this study resolves these discontinuities in the molar relationships at both
SCN8 and SCN12 with an Average Absolute Deviation between the predicted and experimental compo-
sitions of less than 10 percent. This model can quite easily be included in Equation of State packages for
a more accurate description of compositions for Trinidad gas condensates for performing compositional
simulation studies. A partial analysis beyond the C7⫹ fraction is not required with this new model.
Introduction
An accurate description of the compositions of the Single Carbon Number (SCN) groups (pseudo-
components) is an integral part of the reservoir fluids characterization process (Ahmed 1989; Danesh
1998). For gas condensate systems these data are applied with Equations of State (EOS) to evaluate gas
and condensate reserves and production for field development and surface facility design. Very often the
required extended compositional data are unavailable experimentally and are generated from mathemat-
ical models historically known as “splitting schemes”.
2 SPE-169945-MS

Models to extend the composition beyond the measured plus fraction are included in EOS simulation
packages (Ahmed 1984 and 1989; Danesh 1998; Whitson and Brule 2000). The two most extensively used
are the exponential (Pedersen et el. 1984, 1985 and 1989) and the three-parameter gamma distribution
function (Pearson 1895). The use of the latter for this purpose was suggested by Whitson (1983). These
models (Yarborough 1978; Pedersen et al. 1989; Whitson 1983) are applied to gas condensate systems
with the assumption that there is a continuous relationship between pseudo-component composition and
molecular weight. This assumption was based on compositional data limited to North Sea gas condensate
systems and more recently with data from a wider region (Al-Meshari and McCain 2007).
Literature on simulation models recommends using experimental compositional data beyond C7⫹ when
applying extended models to obtain a more accurate description of pseudo-component compositions. It
was not until recently that the limits of the required experimental data were defined for improved
predictions (Hosein and McCain 2009) when using the aforementioned models. These limits were based
on discontinuities observed in the molar relationships at SCN8 and at SCN13 from experimental
compositional data for gas condensates measured with samples with worldwide origin.
In this study, the limitations of the exponential and three-parameter gamma distribution functions were
reviewed and a new model for a more accurate description of pseudo-component compositions is
described for Trinidad gas condensate systems. The initial condensate-gas ratios of the samples tested
were less than 50 stock-tank barrels per million standard cubic feet of separator gas.

Compositional Data Sets used in this Study

A total of twelve sets of compositional data (Appendix A, Table A-1) were used in this study. The
compositions were generated experimentally by a commercial laboratory from separator samples taken at
the various producing gas reservoirs in Trinidad, (Hosein 2004). The compositions of the C7⫹ fractions
were less than 4.0 mole % which would classify these Trinidad gases as lean gas condensates (McCain
1990).

Chromatographic Experiment
Analyses of separator gas and condensate samples were conducted by a commercial laboratory on a
VARIAN gas chromatograph (GC). The instrument was custom designed with sampling valves for
injection, packed columns connected to a thermal conductivity detector (TCD) and a capillary column
connected to a flame ionisation detector (FID) as shown in Figure 1. The arrangement provides the
requirements of GPA Methods 2286-95 and 2186-95 for extended analyses of Natural Gas and Natural
Gas Liquids respectively. The analysis of the separator gas was combined with the analysis of the
separator liquid to yield the composition of the reservoir fluid or well stream (Hosein 2004).

Observed Relationship between Experimental SCN Composition and

Molecular Weight
Yarborough (1978) and Pedersen et al. (1984) observed that plots of mole fraction vs. molecular weight
(or SCN group) for North Sea gas condensates usually exhibit a continuous exponential distribution
function as shown in Fig. 2a or a linear relationship as shown in Fig. 2b (log of mole % vs molecular
weight or semi-log plot). Hosein and McCain (2009) provided experimental data on samples from
worldwide locations that show evidence of distinct discontinuities at SCN8 and SCN13 (Figs. 3a and 3b).
They suggested that these discontinuities should be taken into consideration for a more accurate
description of SCN compositions when using extended models for gas condensates.
SPE-169945-MS 3

Figure 1a—Photograph of Gas Chromatograph Figure 1b—Schematic of Gas Chromatograph

Models tested for extending the Plus Fraction for Gas Condensate
Systems
The Exponential Distribution Function
Pedersen et al. (1984) expressed the observed continuous exponential relationship into a continuous linear
one with a logarithmic expression for mole percent as a function of molecular weight as follows:
(1)

where:
zn ⫽ composition of SCN group n, mole percent
Mn ⫽ molecular weight of SCN group n.
A, B ⫽ constants determined by the least squares fit to the experimental data.
This generally accepted representation of a single straight line relationship is shown in Fig. 2b. With
this model, the Average Absolute Deviation (AAD) obtained between the predicted and experimental
compositions for the twelve data sets is shown in Table A-2 and Fig. A-1. The compositions of the SCN8
groups were under predicted by more than 25 percent whereas the compositions of the SCN12 groups
were over predicted by more than 30 percent due to the discontinuities in the molar relationship at SCN8
and SCN13 (shown in Figs. 3a and 3b). Hosein and McCain (2009) suggested that extended experimental
data up to C20⫹ is required when applying this model. This would provide a minimum of seven
experimental data points to define the discontinuity at SCN13 and beyond as shown in Fig. 3b. Hence this
scheme is more suitable for predicting composition beyond the SCN19 group.
The Three-Parameter Gamma Distribution Function
The three-parameter gamma probability function (Pearson 1895) is used to characterize molar distribution
(i.e. mole percent and molecular weight relation of pseudo components) as follows:
(2)

where: ␣, ␤ and ␩ are parameters defining the distribution (Whitson 1983).

The basic assumption when applying this model to gas condensate systems is also a continuous
(exponential) relation between SCN composition in mole percent and molecular weight (Whitson 1983).
This occurs when the parameter ␣ ⫽ 1. When extending the C7⫹ fraction, the value of ␩ ⫽ 86.177, which
is the molecular weight of n-C6H14 (Al-Meshari and McCain 2007). The Gamma model as described by
Al-Meshari and McCain (2007) was applied to predict SCN compositions for the twelve data sets in Table
4 SPE-169945-MS

Figure 2a—Molar Distribution of SCN Composition for North Sea Figure 2b—Molar Distribution of SCN Composition for North Sea
Sample Sample

(Source: Pedersen et al. 1989) (Source: Pedersen et al. 1989)

A-1. The end points for the integral for each frequency of occurrence calculation used to calculate the
SCN compositions were the molecular weights of the successive normal paraffins.
The AAD obtained between the predicted and experimental compositions of the SCN groups for the
12 data sets are given in Table A-2 and Fig. A-1. These results show that the compositions of the SCN8
groups were under predicted by more than 25 percent whereas the compositions of the SCN12 groups
were over predicted by more than 25 percent, illustrating that this model does not take care of the
discontinuities in the molar relationship at the SCN8 and the SCN13 groups when extending the C7⫹
fraction. Hosein and McCain (2009) demonstrated that extended experimental data up to the C14⫹
fraction are required for best prediction with this model.

Ahmed et al. (1985) two Coefficient Splitting Scheme

The model devised by Ahmed et al. (1985) was based on observation that hydrocarbon systems exhibit
a molar distribution relative to the average molecular weight of the plus fraction. They described a
“marching technique” (see Appendix A) from which the average molecular weights of the plus fractions
(Mn⫹) are calculated from experimental compositional data. They used plots similar to Figs. 4a and 4b to
prepare generalized coefficients (S) for a two segment (two-coefficient) relationship to calculate mole
percent of SCN groups as follows:
S ⫽15.5 for SCN ⫽ 8 and S ⫽ 17.0 for SCN ⬎ 8
The plots were expressed by the generalized equation:
(3)

The Ahmed et al. (1985) splitting scheme was tested using the 12 sets of compositional data from Table
A-1. The most significant observation was that the two segment relationship improved the prediction of
the SCN8 and SCN12 compositions when compared to the exponential and gamma distribution functions
as shown in Table A-2 and Fig. A-1. The AAD between the predicted and experimental compositions of
the SCN8 groups was under 12 percent. However, this value for the SCN12 groups was just over 18
percent. Also there was an over prediction of the composition of the SCN7 group. The Average AAD for
this group was as high as 23 %. This result indicated that the coefficient S ⫽ 15.5 for the SCN8 group
is too high for Trinidad gas condensates.
SPE-169945-MS 5

Figure 3a—Log Mole % vs Molecular Weight for Trinidad Sample PL6 Figure 3b—Log Mole % vs Molecular Weight for Trinidad Sample PL6
(Source: Hosein 2004) (Source: Hosein 2004)

A Proposed “Four Coefficient” Model (4CM) for Splitting the C7ⴙ Fraction
The observed discontinuities at SCN8 and SCN13 and change in slope at SCN13 (Hosein and McCain,
2009) as shown in Figs. 3a and 3b suggest that Ahmed’s splitting scheme should be modified to four
segments and hence four coefficients, instead of two as follows:
1. Segment 1 from SCN7 to SCN8 due to the observed discontinuity at SCN8 and segment 2 from
SCN8 to SCN12 as shown in Figs. 5a and 5b (Hosein 2004).
2. Segment 3 from SCN12 to SCN13 due to the observed discontinuity at SCN13 and segment 4
from SCN13 and beyond as shown in Figs. 6a and 6b (Hosein 2004).
Coefficients for the New 4CM Model
The “marching technique” described in Appendix A was applied to the 12 sets of compositional data given
in Table A-1. The calculated average molecular weights of the plus fractions (Mn⫹) were plotted against
SCN number (7, 8, 9) as shown in Figs. 5a and 5b and Figs. 6a and 6b for sample PL6. The first two
segments shown in Figs. 5a and 5b provide the coefficients S for n ⫽ 8 and for 8 ⬍ n ⱕ 12. The y intercept
is the molecular weight of the C7⫹ fraction. Equation 3 is applied with the “marching technique” to
compute the C7 composition with the coefficient S derived for n ⫽ 8 and compositions from C8 to C11
with the coefficient S derived for 8 ⬍ n ⱕ 12. Predictions from these first two segment relationships (Figs.
5a and 5b) take care of the discontinuity at C8 as shown in Table 1 and Fig. 7 for sample PL6.
(4)

The second two segments shown in Figs. 6a and 6b, which has been mathematically expressed by
Equation 4 from this study, provide the coefficients S for n ⫽ 13 and n ⬎13. The y intercept is the
molecular weight of the SCN12 plus group which is calculated from the “marching technique” as outlined
in Appendix A. Equation 4 is applied with the “marching technique” to compute the composition of
SCN12 with the coefficient S derived for n ⫽ 13 and the compositions of SCN groups greater than SCN12
with the coefficient S derived for n ⬎13. Predictions from these second two segment relationships (Figs.
6a and 6b) for SCN compositions take care of the discontinuity at SCN13 and change in slope from
SCN13 and beyond as shown in Table 1 and Fig. 7 for sample PL6. The deviation between the predicted
and experimental compositions for sample PL6 was less than ⫾8.0 percent as shown in Table 1.
The coefficients S, for each of the four compositional segments described above were determined for
the twelve compositional data sets (Table A-1) and are shown in Table 2. The averages of these
coefficients (shown in Table 3) were applied with the Four Coefficient Model (4CM) described above to
6 SPE-169945-MS

Figure 4a—Molecular Weight of Plus Fraction vs SCN Figure 4b—Molecular Weight of Plus Fraction vs SCN
(Source: Ahmed et al. 1985) (Source: Ahmed et al. 1985)

Figure 5a—Molecular Weight of Plus Fraction vs SCN for Sample PL6 Figure 5b—Molecular Weight of Plus Fraction vs SCN for Sample PL6
(Coefficient S ⴝ 11.0 for n ⴝ 8) (Coefficient, S ⴝ 14.8 for 8 < n <13)

split the C7⫹ fraction of the twelve Trinidad gas condensate samples in Table A-1. The results obtained
from the proposed 4CM model were compared with those obtained by the Ahmed et al. (1985) splitting
scheme (Table 4 and Fig. 8).
Comparison of Results from the Proposed 4CM Model and the Ahmed et
al. (1985) two Coefficient Splitting Scheme
Table 4 and Fig. 8 show that this new 4CM model gives better prediction of SCN compositions than the
Ahmed et al. (1985) two coefficient model. The AAD between predicted and experimental compositions
of the SCN groups was less than 10 percent. It is also important to note that experimental compositional
analysis beyond C7⫹ is not required when applying this 4CM model.
Conclusion
For gas condensates there are discontinuities in the relationship between compositions of the SCN groups
and molecular weights at SCN8 and SCN13. Existing splitting schemes such as the exponential and
gamma distribution function were developed based on the assumption of a continuous relationship. As a
result these schemes under predict the SCN8 compositions and over predict the SCN12 compositions by
more than 25 percent. The use of the gamma probability distribution function does not recognize either
of these discontinuities. The Ahmed et al. (1985) method takes care of the discontinuity at SCN8 but not
at SCN13. Therefore these splitting schemes require experimental extended analysis beyond SCN 14 for
SPE-169945-MS 7

Table 1—Deviation Between Predicted and Experimental SCN Composition for Sample PL6
SCN Group Expt., Mole % Pred., Mole % 4CM Dev., %

7 0.307 0.306 -0.2

8 0.380 0.371 -2.3
9 0.205 0.209 2.0
10 0.157 0.158 0.8
11 0.113 0.121 6.8
12 0.071 0.071 -0.7
13 0.079 0.081 2.5
14 0.062 0.057 -7.3
15 0.053 0.049 -7.0
16 0.039 0.042 7.1
17 0.032 0.034 7.3
18 0.028 0.027 -2.3
19 0.021 0.021 -1.1

Table 2—Coefficients S for each Data Set from the Four Segment Approach
SAMPLE nⴝ8S 8< n <13 S n ⴝ 13 S n >13 S SAMPLE nⴝ8S 8 < n < 13 S n ⴝ 13 S n >13 S

PL1 10.7 15.1 10.1 11.5 CL1 12.1 16.6 11.4 13.3
PL2 12.2 16.1 11.6 13.6 CL2 11.4 15.7 12.1 13.8
PL3 11.4 15.4 11.1 13.3 CL3 11.8 15.8 11.5 13.5
PL4 10.3 14.2 10.3 12.0 CL4 12.2 16.6 10.8 12.5
PL5 10.9 15.6 10.6 12.1 CL5 12.7 17.0 11.5 13.4
PL6 11.0 14.8 10.4 12.3 CL6 10.2 13.9 10.3 11.8

Table 3—Coefficients for Predicting SCN Compositions, S by the “Four Coefficient” Model (4CM)
SCN Number, n nⴝ 8 8< n<13 nⴝ13 n>13
Coefficient Sⴝ11.5 Sⴝ 15.5 Sⴝ11.0 Sⴝ13.0

Table 4 —Average Absolute Deviation between SCN Compositions predicted by Ahmed et al. and the 4CM splitting models and measured SCN
Compositions from Table A-1.
SCN No. of Data AAD, % AHMED et al. AAD, % 4CM

7 12 23.0 8.2
8 12 11.6 6.5
9 12 7.0 7.7
10 12 4.7 4.5
11 12 5.4 6.5
12 12 18.4 8.8
13 12 8.3 6.3
14 12 7.4 5.5
15 12 9.7 6.0
16 12 7.1 8.0
17 12 9.2 7.1
18 12 13.3 7.6
19 12 15.2 8.5

a more accurate description of the compositions of the SCN groups. The proposed new “Four Coefficient”
8 SPE-169945-MS

Figure 6a—Molecular Weight of Plus Fraction vs SCN for Sample PL6 Figure 6b—Molecular Weight of Plus Fraction vs SCN for Sample PL6
(Coefficient S ⴝ 10.4 for n ⴝ 13) (Coefficient, S ⴝ 12.3 for n > 13)

Figure 7—Predicted and Experimental SCN Composition for Sample

PL6

Figure 8 —Average Absolute Deviation between predict and measured

SCN Compositions from Table A-1.

model takes these two discontinuities into account and does not require a partial experimental analysis
beyond C7⫹. This new “Four Coefficient” model can be used to predict the compositions of SCN groups
for Trinidad gas condensates.
Recommendation
The four coefficients generated from this study can be applied to extend the C7⫹ fraction for gas
condensate systems from any region. Improved predictions can be obtained for samples from a particular
region or for multiple samples by generating four new coefficients by applying the equations and method
described for this “Four Coefficient” model.

Acknowledgements
The authors would like to thank the Campus Research and Publication Fund Committee of The University
of the West Indies for providing the financial support for this Research Project.

Nomenclature
AAD ⫽ average absolute deviation
SPE-169945-MS 9

A, B ⫽ constants in Equation 1
bbls ⫽ barrels
C ⫽ Carbon
CGR ⫽ Condensate Gas Ratio
MW ⫽ Molecular Weight
P ⫽ Pressure, psia
PVT ⫽ Pressure, Volume, Temperature
S ⫽ Coefficient
scf ⫽ Standard Cubic Feet
SCN ⫽ Single Carbon Number
SG ⫽ Specific Gravity
UWI ⫽ University of the West Indies
Zn ⫽ mole percent of SCN fraction

Subscript
i ⫽ Component i
n ⫽ SCN fraction number
⫹ ⫽ plus or last fraction

Greek
␣, ␤, ␩ ⫽ parameters in the gamma distribution function.
⌫ ⫽ gamma function
⬎ ⫽ lesser
greaterthan
than

References
Ahmed, T. 1989. Hydrocarbon Phase Behaviour. Gulf Publishing Company, Houston, Texas.
Ahmed, T. H., Cady, G. V. and Story A. L. 1985. A Generalized Correlation for Characterizing the
Hydrocarbon Heavy Fractions. SPE 14266.
Ahmed, T. H., Cady, G. V., Story A. L., Verma, V., and Banerjee, S. 1984. An Accurate Method for
Extending the Analysis of C7⫹. SPE 12916.
Al-Meshari A. A. and McCain W. Jr. 2007. Validation of Splitting the Hydrocarbon Plus Fraction:
First Step in Tuning Equation of State. SPE 104631
Danesh, A. 1998. PVT and Phase Behaviour of Petroleum Reservoir Fluids. Elsevier. Amsterdam
Hosein, R. 2004. Phase Behaviour of Trinidad Gas Condensates. Ph.D. Dissertation. The University
of The West Indies, St Augustine, Trinidad.
Hosein, R. and McCain W. Jr. 2009. Extended Analysis for Gas Condensate Systems. SPE 110152-
PA. Reservoir Evaluation and Engineering, Vol. 12, No. 1, Pgs. 159 –166.
Katz, D. L. and Firoozabadi, A. 1978. Predicting Phase Behaviour of Condensate/Crude Oil Systems
Using Methane Interaction Coefficients. J. Pet. Tech. Vol. 30, No. 11, Pgs.1649 –1655.
Katz, D. L., Hekim, Y. and Firoozabadi, A. 1978. Reservoir Depletion Calculations for Gas Conden-
sates Using Extended Analysesin the Peng-Robinson Equation of State. Canadian J. of Chem. Eng., Vol.
56, Pgs. 1649 –1655.
McCain W. D. 1990. The Properties of Petroleum Fluids, Second Edition, Penn Well Publishing Co.,
Tulsa, Oklahoma
Pearson, K. 1895. Contributions to the Mathematical Theory of Evolution11. Skew Variations in
Homogeneous Material. Philosophical Trans. Royal Society of London, Series A. 186, Pgs. 343–414.
Pedersen, K. S., Thomassen, P., and Fredenslund, Aa. 1984. Thermodynamics of Petroleum Mixtures
10 SPE-169945-MS

Containing Heavy Hydrocarbons. 1. Phase Envelope Calculations by use of the Soave-Redlich-Kwong

Equation of State. Ind. Eng. Chem. Process Des. Dev. 23, Pgs. 163–175.
Pedersen, K. S., Thomassen, P., and Fredenslund, Aa. 1985. Thermodynamics of Petroleum Mixtures
Containing Heavy Hydrocarbons. 3. Efficient Flash Calculation Procedures Using the SRK Equation of
State. Ind. Eng. Chem. Process Des. Dev. 24, Pgs.948 –954
Pedersen, K. S., Fredenslund Aa. and Thomassen, P. 1989. Properties of Oil and Natural Gases. Gulf
Publishing Co., Houston.
Whitson, C. H. 1983 Characterizing Hydrocarbon Plus Fractions. Soc. Pet. Eng. J., Vol. 23, No. 4
683–694.
Whitson, C. H. and Brule, M. R. 2000. Phase Behaviour, Monograph Series, SPE, Richardson, Texas
20: 35
Whitson, C. H. and Torp, S. B. 1983. Evaluating Constant Volume Depletion Data. J. Pet. Tech.:
610 –620.
Yarborough, L. 1978. Application of a Generalized Equation of State to Petroleum Reservoir Fluids,
Amer. Chem. Soc. 182: 386 –439.
SPE-169945-MS 11

Appendix A
Splitting Scheme of Ahmed et al. (1985) – “Marching Technique”
The “marching technique” devised by Ahmed et al. (1985) was based on observation that hydrocarbon systems exhibit a molar
distribution that is relative to the average molecular weight and specific gravity of the plus-fraction. It involves calculating the
composition zn at a progressively higher SCN fraction as follows:
A-1

where:
zn ⫽mole percent of the extended SCN fraction
MWn ⫽molecular weight of the SCN fraction as outlined by Katz and Firoozabadi (1978).
MWn⫹ ⫽ molecular weight of the n plus fraction (C8⫹, C9⫹. . .. . .), as calculated by the following expression:
A-2

Where the subscript n is the number of the SCN fraction and the coefficient S for gas condensate systems ⫽ 15.5 for n ⫽
8 and 17.0 for n ⬎ 8.
MW (n ⫹ 1)⫹ in Equation A-1 is the molecular weight of the next plus fraction and is also calculated from Equation A-2,
but with the next n value, i.e. n ⫽ n ⫹ 1.
Procedures for obtaining molecular weight and specific gravity of the n⫹ fractions, i.e., MW n⫹ and SG n⫹ were
summarized by Ahmed et al. as follows:

1) Given the composition; specific gravity; and molecular weight for a hydrocarbon system, up to C7⫹, all components
heavier than C7 are grouped into a “plus” fraction, i.e. C 8⫹, which is characterized by an average molecular weight
MW8⫹, a specific gravity SG8⫹ and a total mole percent z8⫹.
2) The average molecular weight and specific gravity of C8⫹ are than calculated from the following relationship:
A-4

and
A-5

where
MW7⫹, SG7⫹ ⫽ measured molecular weight and specific gravity of heptanes plus respectively
MW8⫹, SG8⫹ ⫽ calculated molecular weight and specific gravity of octanes plus respectively
MW7, SG7 ⫽ average molecular weight and average specific gravity of heptanes as recommended by Katz-Firoozabadi
(1978)
3) The physical properties of the next hydrocarbon “plus” fraction i.e. C9⫹ are calculated following the procedures outlined
in steps 1 and 2, thus,
(A-6)

and
(A-7)

This “marching technique” for calculating average molecular weight and specific gravity is repeated until the last
component in the hydrocarbon system is reached.
Average Absolute Deviation, AAD and Deviation, Dev
Average Absolute Deviation, AAD ⫽

Deviation, Dev. ⫽
12 SPE-169945-MS

Table A-1—Compositional Data and Properties of C7ⴙ for the Twelve Trinidad Gas Condensate Samples Used in This Study
SCN PL1 PL2 PL3 PL4 PL5 PL6 CL1 CL2 CL3 CL4 CL5 CL6
Group Mole % Mole % Mole % Mole % Mole % Mole % Mole % Mole % Mole % Mole % Mole % Mole %

C7 0.561 0.458 0.508 0.346 0.368 0.307 0.588 0.353 0.390 0.523 0.414 0.294
C8 0.789 0.583 0.652 0.444 0.471 0.380 0.729 0.457 0.486 0.633 0.504 0.367
C9 0.491 0.319 0.353 0.308 0.297 0.205 0.491 0.304 0.322 0.352 0.310 0.234
C10 0.354 0.245 0.286 0.231 0.208 0.157 0.327 0.222 0.225 0.238 0.215 0.159
C11 0.267 0.196 0.194 0.165 0.141 0.113 0.246 0.162 0.171 0.196 0.162 0.112
C12 0.176 0.127 0.134 0.108 0.089 0.071 0.156 0.118 0.115 0.110 0.100 0.070
C13 0.197 0.148 0.146 0.122 0.101 0.079 0.169 0.129 0.123 0.123 0.112 0.080
C14 0.170 0.105 0.121 0.095 0.079 0.062 0.151 0.101 0.106 0.099 0.090 0.056
C15 0.144 0.093 0.103 0.084 0.069 0.053 0.121 0.088 0.093 0.086 0.079 0.046
C16 0.119 0.074 0.077 0.068 0.052 0.039 0.098 0.068 0.069 0.066 0.064 0.039
C17 0.104 0.066 0.065 0.059 0.045 0.032 0.085 0.059 0.059 0.055 0.056 0.032
C18 0.099 0.054 0.052 0.049 0.037 0.028 0.076 0.052 0.054 0.047 0.047 0.027
C19 0.080 0.045 0.041 0.040 0.031 0.021 0.064 0.044 0.043 0.039 0.039 0.019
C20⫹ 0.373 0.240 0.187 0.149 0.134 0.069 0.344 0.232 0.222 0.184 0.218 0.057
C7⫹ 3.924 2.753 2.919 2.268 2.122 1.616 3.645 2.389 2.478 2.751 2.41 1.592
SG7⫹ 0.8031 0.8004 0.7939 0.7967 0.7918 0.7869 0.8021 0.8049 0.8022 0.7925 0.8005 0.7849
MW7⫹ 160 157 150 153 148 143 159 162 159 148 157 141

Table A-2—Average Absolute Deviation between SCN Compositions predicted by the Exponential, Gamma and Ahmed et al. Models and mea-
sured SCN Compositions for the 12 Samples in Table A-1.
SCN Group No. of Data AAD, % EXPON. AAD, % GAMMA AAD, % AHMED et al.

7 12 4.5 8.7 23.0

8 12 26.2 26.8 11.6
9 12 8.7 10.6 7.0
10 12 4.2 6.4 4.7
11 12 14.1 14.2 5.4
12 12 35.1 31.6 18.4
13 12 7.8 11.3 8.3
14 12 7.6 13.6 7.4
15 12 3.2 10.9 9.7
16 12 3.1 19.0 7.1
17 12 4.9 15.5 9.2
18 12 9.9 9.5 13.3
19 12 8.2 8.1 15.2

Figure A-1—Average Absolute Deviation among predicted and mea-

sured SCN Compositions for the 12 Samples in Table A-1.